Multifunctional design from disordered solids to self-assembly

Despite the increasing ability to control forces on the sub-micron scale in a wide variety of systems, knowing how to set interactions to obtain a desired outcome remains a considerable challenge. Of increasing importance to self-assembly and material design is the extreme where there are many control knobs as well as many features that one wishes to simultaneously control. Can we predict what target features can be tuned independently, or the susceptibility of certain feature combinations to various parameters? I will discuss how a simple linear theory, enabled by differentiable programming, presents a unified framework for understanding such problems, allowing us to directly answer such questions. Using examples from designing bulk properties of disordered solids to the design of self-assembling structures and their kinetics, I will show how this approach provides the language and tools necessary for understanding a wide and diverse range of complex inverse problems.